Machine learning assesses drivers of PM_2.5 air pollution trend in the Tibetan Plateau from 2015 to 2022

The Tibetan Plateau (known as the Earth's Third Pole) has significant impact on climate. Fine particulate matter (PM_2.5) is an important air pollutant in this region and has significant impact on health and climate. To mitigate PM_2.5 air pollution over China, a series of clean air actions has been implemented. However, interannual trends in particulate air pollution and its response to anthropogenic emissions in the Tibetan Plateau are poorly understood. Here, we applied a random forest (RF) algorithm to quantify drivers of PM_2.5 trends in six cities of the Tibetan Plateau from 2015 to 2022. The decreasing trends (-5.31 to -0.73 μg m^-3 a^-1) in PM_2.5 during 2015-2022 were observed in all cities. The RF weather-normalized PM_2.5 trends - which were driven by anthropogenic emissions - were -4.19 to -0.56 μg m^-3 a^-1, resulting in dominant contributions (65 %-83 %) to the observed PM_2.5 trends. Relative to 2015, such anthropogenic emission driver was estimated to contribute -27.12 to -3.16 μg m^-3 to declines in PM_2.5 concentrations in 2022. However, the interannual changes in meteorological conditions only made a small contribution to the trends in PM_2.5 concentrations. Potential source analysis suggested biomass burning from local residential sector and/or long-range transports originated from South Asia could significantly promote PM_2.5 air pollution in this region. Based on health-risk air quality index (HAQI) assessment, the HAQI value was decreased by 15 %-76 % between 2015 and 2022 in these cities, with significant contributions (47 %-93 %) from anthropogenic emission abatements. Indeed, relative contribution of PM_2.5 to the HAQI was decreased from 16 %-30 % to 11 %-18 %, while increasing and significant contribution from ozone was observed, highlighting that further effective mitigation of both PM_2.5 and ozone air pollution could obtain more substantial health benefits in the Tibetan Plateau.